Subscribe
Share
Share
Embed
upGrad Enterprise aims to build the world’s largest GenAI learning initiative to enable high-growth companies to embrace technology’s transformative business impact. Hosted by Srikanth Iyengar, CEO, upGrad Enterprise, the GenAIrous Podcast, will curate an exciting roster of global experts and guests, who are at the cutting-edge of Generative AI, and its varied applications in the world of business.
Welcome to the GenAIrous Podcast where we unravel the fascinating world of generative AI and its transformative impact on business globally. I'm your host, Srikanth Iyengar, CEO of upGrad Enterprise. At upGrad Enterprise, we're building the world's largest GenAI learning initiative, empowering high growth companies to leverage cutting edge technology. Each week, join me and the roster of global experts as we explore innovations shaping the world of work as we know it. Let's get GenAIrous.
Srikanth Iyengar, CEO upGrad Enterprise:Hello, and welcome to another episode of the GenAIrous Podcast. It's such a pleasure today to have R Arun Kumar or Rak with us. Rak, good to have you here. I see you're in sunny California.
Arun Ramchandran, President - Hexaware:I'm delighted delighted to be part of this podcast, Srikanth.
Srikanth Iyengar, CEO upGrad Enterprise:Wonderful. Thank you so much. And, you know, we've had some interesting guests in the last few episodes. I'm sure this promise is to be that and even more. So really, really excited about the chat. So, Rak, I mean, look, first of all, you work in a very exciting space with a very exciting fast growth company, Hexaware, and I know that, you know, you're part of the core senior leadership team that is helping reinvent this company into a completely different avatar, AI first, technology first. So any thoughts on that? How are you guys going about it, and why that particular thesis?
Arun Ramchandran, President - Hexaware:Yeah. No, I'm delighted, to share a little bit more about it. Podcast is my favorite medium, by the way. So I'm I'm enjoying being part of this. Thanks for having me. So, Hexaware is a So Hexaware is a Carlyle owned company, but we were not always Carlyle owned. We were owned by another private equity before that. And then we were public, in the markets. We have had a couple of pivots. Right?
Arun Ramchandran, President - Hexaware:So 10 years ago, we pivoted hard towards what I would call as automation. We picked up the cloud pivot a couple of years later. We saw this trend coming. So our our whole focus has been looking ahead and seeing what technology trend is gonna most impact the enterprises and our business model, and essentially putting in capabilities, investing in it, getting trained, getting ready. I think a few years ago, the whole AI and data part was, you know, obviously becoming important.
Arun Ramchandran, President - Hexaware:We decided to, build capabilities around it. What we're seeing, especially in the last 2 years; one and a half, 2 years is after Carlyle came on board, that our ability to now think big and actually pivot towards an AI first kind of a model has become much more real. That is really what we are doing now. We are looking at transforming every part of our our own culture, our own operations into an AI driven, model. Whether it is the services that we deliver to our clients across engineering, across testing, across IT operations, whether it is the kind of, solutions that we are taking using our platforms, RapidX, 10X AI, Amaze, our own decision making, our own employee engagement. 90% of our workforce, technology workforce is now trained on AI and Gen AI fundamentals, and it's complete continues to to, upscale based on that. So we are looking at it helping with our own service delivery model because I think in the future, no business model will be untouched by AI.
Srikanth Iyengar, CEO upGrad Enterprise:I remember you mentioning in a previous discussion we had that it's also opening new industries or new markets for you as well. Any any thoughts on that as to how this could be a huge revenue spinner for companies like you?
Arun Ramchandran, President - Hexaware:Yeah. So the way we see this, Srikanth, and, you know, our association on these kind of transformation goes back long ways. Right? So we've seen how these waves can impact not just the actual business model, but also growth markets. The way we have starting to see this, this will lead to a service model redefinition. It will lead to fundamental change in the economics. Instead of moving into a resource focused model, time and material, this will move towards monetizing of platforms and monetizing of technology and data. The way cloud was important a few years ago, data is becoming very critical for our customers as well, and we see that. How do you unlock that data and how do you actually create future business value exponentially using that hidden data is gonna be one of the promises from AI and Gen AI. And that is the growth market that we see across different industries.
Arun Ramchandran, President - Hexaware:Some industries are gonna be early adopters, information intensive industries like financial services, health care. You know, these are all very, very focused on data and data flow, information flow. I think the the kind of business model impact this could have will be huge. There are some regulated industries which will take this on once those regulatory concerns are sorted out. Right? But it's gonna flow through the all the industry sectors. McKinsey published a study, I think, one and a half years ago which talked about 1,000,000,000 of dollars of upside, right, from all the the different parts of the value chain which are gonna get impacted. And a lot of it is based on data and, unlocking of that data.
Srikanth Iyengar, CEO upGrad Enterprise:No. Absolutely. Couldn't agree more. And, like you said, there have been many, many reports. McKinsey is one such, but by a few other consulting firms, governmental bodies, research firms that talk about this being an inflection point almost akin to the Internet, if not more so. So couldn't couldn't agree with you more. But but we're just shifting tracks a bit on that track. Clearly, you know, the generative AI as a trend is relatively new, but AI has been around for decades. Right? I mean and generative AI is one part of what one would call the AI in the machine learning space. So do you wanna shed a bit of light on that evolution for our listeners?
Arun Ramchandran, President - Hexaware:Yes. I think everybody has heard the word AI, but till generative AI came on board, AI used to be a term at a very esoteric level, mostly a boardroom kind of discussion maybe, technology, ivory towers, research and academia. I think Gen AI has made it more of a kitchen table conversation. You know? For a lot of people it's democratized the aspect of leveraging AI for solving business problems and day to day life issues. So you're right. It's thrust into the limelight. I think the evolution we are seeing more from a business perspective. Right? I talked about automation.
Arun Ramchandran, President - Hexaware:I think traditional AI machine learning was a lot about predictive analytics, right, And it included a huge amount of automation. I mean, you understand if you look at the the actual mechanics behind it, the neural networks, the deep neural networks, mimicking a little bit of the human brain, you know, looking at nodes, looking at weights, looking at parameters, that has existed. That particular technology, the architecture has existed. What has changed is the whole transformer part to it and how it mimics human brains in terms of learning and outputting. But that is not enough, I think what is happening now is it's moving from that automation phase to augmenting human capabilities.
Arun Ramchandran, President - Hexaware:Right. So we look at it as a 4 A's kind of a approach. Automation was the first phase, Augmenting of human capabilities, which is can you know, your constraint from for time. It can summarize something for you. You want some ideas on design, it can come out with it. You want to, you know, respond to somebody quickly, you know, through a voice to, text transcription, it can do that. So there's a lot of augmentation. The promise now that we are starting to see which is gonna impact the business models is the amplification. What do I mean by that? Processing powers are huge.
Arun Ramchandran, President - Hexaware:Ability to crunch data, I mean it can read through 1,000,000,000 of dollars or 1,000,000,000 of words which human beings in their lifetime don't do. Right? So it can generate, see through a lot more patterns and come out with output in a faster timeframe and which is what all these measures are, it can amplify human capabilities hugely. And at some point in time, starting to happen in small corners, it can even work autonomously. So right now it's working with a lot of human interaction, human guidance, but as it becomes more safe, more trustworthy, it could take on more autonomous role.
Arun Ramchandran, President - Hexaware:So we look at it as the 4A phase moving from a traditional predictive AI to more of a generative AI and what does that really mean in terms of value? It's based on automation, augmentation, amplification, and autonomy.
Srikanth Iyengar, CEO upGrad Enterprise:No. That's a very succinct, 4A framework. Thank you for that.
Arun Ramchandran, President - Hexaware:Yeah. No. I'll share I'll share if you don't mind, I can share a personal anecdote on it. Please, please,
Srikanth Iyengar, CEO upGrad Enterprise:Go for it.
Arun Ramchandran, President - Hexaware:Chat GPT came out, and I've I've been in the technology world, and I sometimes experiment with this. But in this case, my wife who's a teacher, she actually created a Chat GPT account before I ever did, and she started experimenting with it because it's just so easy, so intuitive, and so helpful in the kind of jobs, you know, in the kind of role that she is doing. So I just wanted to say how quickly it has made inroads into people who may not otherwise be very friendly to new technologies right away.
Srikanth Iyengar, CEO upGrad Enterprise:Look. Completely. And I can share an example as well. You know, I've seen the debate firsthand. As you know, I have a teenage daughter, and her school had a live debate of whether they can or not use Chat GPT to support them with school assignments. And, of course, there was the aspect of moving through it quickly, they and they debated whether they learn enough or not and whether it's ethical or not. And to see 16 year olds have that discussion and have very opposing points of view was fascinating. I have never seen her as fascinated by an underlying technology as I've seen her do with this. So completely completely agree with you. So thank you for that.
Arun Ramchandran, President - Hexaware:We are in a very exciting phase of, technology and you mentioned it. Right? The way internet disrupted it. I think some of the disruption here we cannot even foresee. It's gonna happen a few years from now. Googles and the Amazons of the world came into play 6 or 7 years after internet became mainstream. So that is the kind of you know, maybe a delay factor we are talking about after the hype cycle subsides, after the infrastructure investments have happened. Who remembers WorldCom? But they are the ones who put the fiber optics in place for creating a lot of the internet infrastructure and then the Amazons and Googles rode on top of it. I think the same thing is gonna happen here.
Srikanth Iyengar, CEO upGrad Enterprise:So let's go since you're talking about enterprises that have won the race so to speak, and adopted this new technology, let's talk about enterprises today. Clearly, this is a debate in the boardroom. This is a debate with all leadership teams. This is a debate with mid level execs. It's something that, you know, people at all as aspects of an organization are discussing. Use cases are being discovered. They are being prioritized. In some cases, they are being implemented as well. And you have a very, very unique vantage point because Hexaware works with global companies across industries, and across different parts of their value chain. So any key use cases, any sort of examples you'd like to highlight that bring to bear the potential of this technology?
Arun Ramchandran, President - Hexaware:Yeah. Yeah. And a lot of it is still evolving, Srikanth. You know, we thought and everybody thought in this industry that last year 2023 was gonna be the year of the POC, you know, proofs of concepts and pilots. And this year 2024 was going to be the year of production. Arguably, that has not really happened or transpired to the extent people thought it would. Lot of factors behind it. I mean the POCs are going up, the use cases are mushrooming, we see a lot of experimentation. It's like dating. People are going out on dates, trying to check this out, this model, this use case but the long term commitment part is still few and far between.
Arun Ramchandran, President - Hexaware:So I just wanted to preface this. Right? You see that in the investment side too. So just taking Microsoft as an example. So there is a gap between enterprise adoption. Lot of CIOs, lot of CTOs are saying yes, they do want to adopt. So if you look at their intent, 85%, 90% think it will be part of their their, you know, business, their enterprise tech stack. But only 15-20 percent have actually done anything at an enterprise level. Almost everybody is experimenting. And we are seeing the same.
Arun Ramchandran, President - Hexaware:So we have over 70% of our top 100 clients discussing, having conversations of different stages on Genai. A few of them we are actually taken forward for production level. And I'll share with you some of the examples. For an insurance company, especially in Europe, we have 400 agents actually using GenAI based contact center for responding to customer queries. This is in the P&C business.
Arun Ramchandran, President - Hexaware:So they have customers calling them about coverage, about moving from one jurisdiction to the other while still keeping GDPR compliance aspects, you know, in mind. So how we have worked on the different kind of datasets and data sources, making sure that this is not being this is kept confidential, not being used for training the core models, the different kind of prompting which is needed, wise to text transcription on a real time basis so that they can advise clients on, you know, hey. You're missing on this coverage or maybe you should look out for this particular item in your policy. So that is a real production kind of use case that we are doing. We are seeing some use cases on the software development life cycle.
Arun Ramchandran, President - Hexaware:Right? So right in terms of coding and engineering, code comprehension, refactoring of a legacy code, creation of automated test scripts. Those are another set of use cases where there is a lot of production ready kind of deployments happening. So I did mention operational productivity. I did mention customer care. I did mention knowledge management and then software development life cycle.
Arun Ramchandran, President - Hexaware:There are 1 or 2 which are very interesting on the domain side. There is a hardware reseller which has us which is working with us to create product descriptions of their very hard to define kind of SKUs and making sure that GenAI is used for making it more search engine optimized and better to be found in ecommerce catalogs. And their accuracy, product accuracy and search findings and sales have actually improved. We have seen that in real life. So many such many such engagements going on across different world.
Srikanth Iyengar, CEO upGrad Enterprise:No. Those are fantastic examples because you've touched different industries. One of the big constraints that we hear about to large scale adoption or enterprise adoption are the risks around generative AI. Data privacy, like you said, there's regulation in the EU. There's regulation in the US. There are concerns about whether this is proprietary to a company and if they use an, a sort of an open model. What does that mean for their, let's say, secret sauce, if you call it that? There are concerns about employees, misusing it. So what is your take on all this? Because these are actual issues that clients are grappling with on a daily basis.
Arun Ramchandran, President - Hexaware:They are. And that's one of the reasons what is holding back large scale adoption. And some of them are real. Some of them, there have been improvements in technology state of the art in the last several months to mitigate the potential risks. One of the main risks that people talk about is hallucination. I mean, it's a term which is used for the Gen AI model coming out with inaccurate data or wrong information or wrong output, albeit in a very confident manner. Okay. So you don't realize that there is actually a little bit of inaccuracy there or sometimes it's just plain wrong. There are techniques both in terms of model selection and model pre training. There are techniques in terms of the right kind of prompt engineering.
Arun Ramchandran, President - Hexaware:There is the whole technology around retrieval augmented generation where you use your own enterprise dataset to feed the right answers before it goes to the model. There is Red Teaming. There is a real life human feedback. So there are aspects like that to prevent or reduce hallucination.
Arun Ramchandran, President - Hexaware:You talked about security. There are, again, multiple areas in which security can be compromised. But it's not just in Gen AI. You see that in IT in general, whether it's cyberattacks and different kinds of security breaches. That's an ongoing challenge that the technology world had has with hackers and people who want to exploit vulnerabilities. That will continue to be there.
Arun Ramchandran, President - Hexaware:But within the Gen AI realm, ability to keep your data confidential, to keep your data within the enterprise, not share it for training the models, ability to detect prompt injection attacks and be able to cut it, ability to do what is called as Red Teaming. Security breaches can happen at the data side, at the model side, and in the integration endpoints. Right? So how the red teams can actually figure out where the vulnerability is, and there are different names to it. It could be data poisoning. It could be prompt injections. It could be, you know, I've mentioned the data, the the kind of integration, you know, hacking that can be done. So all of them can be managed. There is a regulatory framework in place also, which which is making sure that the model providers have certain specific guardrails in place. And and, you know, whether it's Microsoft, whether it's AWS, they have all made commitments about not using the enterprise data, client data to train their models, providing unlimited liability or backstop to any claims, 3rd party claims, which might arise.
Arun Ramchandran, President - Hexaware:So all that is getting in place. I think there is a regulatory framework emerging, which should help with this.
Srikanth Iyengar, CEO upGrad Enterprise:I think like most of the technologies, Rak, there will be safety guardrails that will come up, but people will solve for them as we go forward. There were concerns if we, again, go back to the Internet as an example. There were concerns about paying on the Internet, but that's how we pay today. There were concerns about tap payment, touchless payment, but we've figured out a way to do that. So clearly, it's it's a function of, I think, humans getting comfortable with the use of the technology and potential, let's say, risks or gaps being found and being solved for, and that's an ongoing process.
Srikanth Iyengar, CEO upGrad Enterprise:But I think it's fair to say that the potential of the technology views widely is probably far more attractive than the downsides. I think that's at least the way one would look at it, I would think. The other aspect that's being discussed a lot is the sustainability aspect, the impact on the environment, because we talked a lot about the data that's being consumed by these models, and that also requires a lot of compute power. And then that consumes a lot of energy, which is impacting the environment, and that's a concern that enterprises and individuals have. So how do you see companies dealing with that?
Arun Ramchandran, President - Hexaware:Fascinating question, Srikanth. Glad you brought it up because right now, the kind of electricity consumption because of these AI models is really going up. In fact, even more than the blockchains and so on. Right? There are certain technology developments which are happening. Each of these models is based on intensive computation. Those computation happen based on a term called flops. Right? Floating point operation points. There is a certain level of accuracy which these models have in their nodes or weights and parameters. By just reducing that set of decimal points which are needed to calculate and keeping the accuracy level a little bit on the wider tolerance level, the consumption, the computing power requirement, and the electricity consumption can come down quite a bit. But, firms like Nvidia and some others are also looking at cooling together the various processing units. Right now, they you know, the whole GPU concept is based on several parallel processing, and hence, the kind of electricity which needs to flow through it, the kind of cooling requirements and so on. So there are companies there is a company called Coreweave, which is looking at centralizing and putting all of them under 1 on one chip.
Arun Ramchandran, President - Hexaware:Right? It could be a 3D kind of an architecture, which could help reduce aspects of that. Just by using better programming, and you've all heard of the CUDA software which Nvidia uses, There are other companies which are starting to use that. Companies like, you know, Google and Microsoft are coming out with their own chips, which will be more integrated in terms of the software which needs to program these chips as the training happens and as the inferencing happens. A lot of money has gone into training but we see more and more inferencing happening which is when the models are actually getting used.
Arun Ramchandran, President - Hexaware:How do you make that inferencing itself less costly and less computation intensive? That is through smaller models, more purpose built kind of integrated, let's say, inferencing and training kind of, environments. So there are multiple things which are starting to happen just to look at how to reduce and create a more sustainable future for AI. Even there is what is known as a post transformer architecture which is coming into play where the whole transformer architecture is based on a self attention mechanism. What it means is that every time you have a new request or a new prompt, everything in that particular set of parameters or that, network is getting recomputed because it's seeking attention and then it sort of sort of go goes exponentially.
Arun Ramchandran, President - Hexaware:There is a new form of architecture post transformer which reduces the need, there is a more linear part to it. It reduces the exponentiality of additional computation just because you're asking more questions and the more prompts are coming in. So I think things are being done, hardware, software prompt, architecture levels to make it more sustainable. People are becoming more, aware of the the requirements here. Let's let's keep fingers crossed and hope that, this also leads to something which doesn't which which actually, you know, doesn't lead lead to environmental, you know, harms.
Srikanth Iyengar, CEO upGrad Enterprise:So completely, I think, those points are super important. Like you said, GPU design, cloud compute, the architecture itself, CUDA, because, you know, we need to harness this technology because while on one side, it is impactful in the environment, if not used wisely and, you know, but the on the other side is helping solve some of the biggest problems that humanity faces. So I think, like, we've had a fascinating discussion, but I do want to make it, a little more personal for many of our listeners. As you know, upGrad we we skill a few million people globally. Many of them are quite young at the early stage of their career. You're someone who's straddled the world of business and technology at various senior levels now for 3 decades. They're at the start of where they are at. Given the this inflection point, what would your advice to them be?
Arun Ramchandran, President - Hexaware:Srikanth, I think the question is actually should be the opposite way around. These folks, young folks are more technology savvy and this is more intuitive to them. Like, you heard the term digital natives, you know, which became which came of the age in the era of Internet and mobile and so on. Can't even they can't even think of a world world where people are not connected all the time and information is not available on fingertips. I think now we you know, the Gen AI word is self refer referral in that sense. The new generation is gen AI generation. You know, just like we have Gen X and Gen Y, this is a Gen AI generation. I mean, they know this intuitively. They will use it. And a few years from now, they will be like, what? We didn't have agents to do this kind of thing before? We had to do it on our own? I mean, a lot of people who drive nowadays, they don't look at maps, paper maps. Right? They're so used to GPS and digital stuff and so on.
Arun Ramchandran, President - Hexaware:I think the same thing will be with Gen AI agents. It will become part and parcel. But jokes aside, I think as I was saying, the the youth of today, the gen AI as may call it, the gen AI generation understands a lot of this intuitively. But if at all there is advice I can give is experiment with it. Right? This is easily accessible. It's part and parcel of your regular software, regular office suite. If you're working in an office environment, you most likely have access to a GPT. You have, you know, people who are using Microsoft Suites. They will probably have access to different kind of copilots.
Arun Ramchandran, President - Hexaware:You can use Gen AI on day to day. You can use it to summarize your emails. You know, you all have experienced that dread when you wake up in the morning and you have these 10 or 15 threats to the same email. Somebody sent an email out, somebody else has replied, 2 other people have commented on it, there are more and then suddenly there are 10, 15 emails on the same topic and you're looking at it and say, oh my god. I had to go through right from the bottom. You can actually use your Copilot to summarize it quickly, and it will come back with this person said this, this person said this, these are the open items, these are the action items, this is what the summary is. It'll easily improve your productivity in that manner. We are using Gen AI for case study creation. Right? We're looking at marketing content.
Arun Ramchandran, President - Hexaware:Somebody has written up some stuff. There are 2, 3 different presentations. Can you combine it into a short nifty case study? How many of us have faced that problem where we have to go for a presentation and we are scrambling to create a case study out of a bunch of information? So these are very small, easy to implement use cases that you can use Gen AI for apart from what you're doing in your personal life, like maybe coming up with a new creative recipe to cook at home.
Arun Ramchandran, President - Hexaware:Back to a kitchen table conversation.
Srikanth Iyengar, CEO upGrad Enterprise:Back to the kitchen table. And that brings me to one more question, Arun. Actually, question I'll have for you is flipping it around. A lot of the people who are making decisions on this today are people who are senior, advanced in their careers in various companies. So what would you say to them? How do they acquaint themselves with this technology? Because they are making decisions that are gonna impact lots of people sort of, you know, and lots of companies. And, what would you say to them?
Arun Ramchandran, President - Hexaware:Either lead the way or get out of the way. Right? But, you know, we've been having so many conversations with senior stakeholders. I have not seen a single person who is not aware of the impact this technology is going to have or could have. So they are taking it seriously. There are some cynics obviously, there is a hype cycle and so on. People want to learn. They want their teams to learn. Like I did a workshop last week with a very senior management team of an investment bank firm. This is the business transformation team.
Arun Ramchandran, President - Hexaware:We had done something with the technology team and saying look actually you know what you guys did something for the technology team. It was very intensive from a tech perspective. We want to know more from a business side. What does all these things mean? How do we handle it? How we roll it out? What is chain management? What kind of skill sets are needed? So people are paying attention, people are getting ready for this change. As I said in the right very beginning, becoming an AI first is not just gonna be our imperative.
Arun Ramchandran, President - Hexaware:We believe every organization is going to start looking at how AI is gonna change their business model both for their own employees and for their customers and what it means in terms of future economics, service delivery, ways of working, and the kind of people which which are hired. So I think it's gonna be pervasive.
Srikanth Iyengar, CEO upGrad Enterprise:No. Super. Rak, thank you so much. This has been a fascinating discussion. I think we've touched on a lot of things from usage in one's personal life to deep technology to enterprise usage to the age spectrum and how different people adopt technology.
Srikanth Iyengar, CEO upGrad Enterprise:So, you know, very, very wide ranging discussion. Really appreciate it. And once again, thank you, for your your support. We are glad. Upgrad is really glad to be partnering with Hexaware and what promises to be a very exciting journey over the next few years.
Arun Ramchandran, President - Hexaware:And I've been delighted. This was a great conversation, Srikanth. You asked some very perceptive, very wide ranging set of questions. And I think the journey we are all on together as partners, as colleagues, I think looking forward to how what the future brings, but excited about it.
Srikanth Iyengar, CEO upGrad Enterprise:Wonderful. Likewise. Likewise. Can't wait. Let the ball start rolling as they say. Super. True.
Srikanth Iyengar, CEO upGrad Enterprise:And that concludes another episode of the GenAIrous Podcast. We are very grateful to our guests for their time and expertise. A big thank you to our producer, Shantha Shankar in Delhi, and our audio engineer, Nitin Shams in Berlin for making magic happen behind the scenes. Join us next time, and don't forget to subscribe to GenAIrous wherever you listen to your podcasts.